YourMemory — Persistent Memory for AI Agents | MCP Compatible
MCP COMPATIBLE 89.4% RECALL@5

Your AI should
remember you.

YourMemory gives any AI assistant persistent, intelligent memory across sessions — with biologically-inspired decay that forgets the right things at the right time. Open source. Runs entirely on your machine.

Get started free
$pip install yourmemory
89.4%
Recall@5 · LongMemEval
−84%
Token waste · 30 sessions
7
MCP clients
2 cmds
To full setup
// WRITTEN ABOUT BY
{{ m }}
// HOW IT WORKS[ 01 / 07 ]

Memory that works like yours.

Three steps. Your AI agent learns, remembers, and naturally forgets what it no longer needs.

{{ step.n }}

{{ step.title }}

{{ step.desc }}

{{ step.code }}
// THE ENGINE[ 02 / 07 ]

Three layers. One engine.

Vector search finds what you asked for. The graph finds what you forgot to ask for. Ebbinghaus decides what survives.

{{ layer.tag }}

{{ layer.title }}

{{ layer.desc }}

{{ pt }}
// MEMORY DECAY[ 03 / 07 ]

Watch memory fade.

Every memory carries a strength score S(t) = e−λt. Different categories age at different rates. Drag the half-life — when strength drops below the prune line, the memory fades from retrieval.

half-life{{ halfLifeLabel }}
{{ strengthLabel }}
strength @ 14d
{{ pruneDayLabel }}
pruned at S < 0.05
// BENCHMARKS[ 04 / 07 ]

Measured, not claimed.

Evaluated on three public datasets. Every number independently reproducible — benchmark code lives in the repo.

{{ bigNum }}
{{ bigLabel }}

{{ bigDesc }}

{{ bar.l }} {{ bar.valLabel }}
// YOUR NUMBERS

How much would you save?

Two ways to see your numbers — pick whichever fits.

1
Instant audit — runs entirely in your AI session

Copy this prompt and paste it at the end of any Claude, ChatGPT, or Cursor session. Your AI will reply with the exact token breakdown and a compressed memory block — no need to paste anything back here.

{{ auditPrompt }}
Your AI will respond with: Full session: ~X,XXX tokens · YourMemory would store: ~XXX tokens · Savings: XX%
or calculate from your full history
2
Deep dive — paste your full conversation history

Export or copy your full session transcript and paste it below. We tokenize it locally and show you the cumulative cost across sessions.

{{ calcTokensFmt }} {{ calcTokensLabel }} ~4 characters per token
120
How many separate chats does this history cover?
Paste your full conversation history,
set sessions & model, then click Calculate savings.
No text pasted yet
Paste your full conversation history in the textarea on the left, then click Calculate savings again.
Without YourMemory
{{ calcBaseTokFmt }}
tokens total · {{ calcBaseCostFmt }} on {{ calcModelLabel }}
Context grows each session — by session {{ calcSessions }}, you're re-sending {{ calcCtxPerSessFmt }} tokens of history every time.
With YourMemory
{{ calcYmTokFmt }}
tokens total · {{ calcYmCostFmt }} on {{ calcModelLabel }}
~{{ calcYmPerSessFmt }} recalled tokens injected per session — flat, never growing.
You save
{{ calcSavePct }}%
{{ calcSaveTokFmt }} tokens · {{ calcSaveCostFmt }}
// COMPARISON[ 05 / 07 ]

Not a vector database.

Vector stores hoard chunks and never forget. YourMemory keeps memory that strengthens, fades, and connects — like yours.

Legacy · vector DB

Stores chunks. Returns chunks. Each session starts from zero — and nothing is ever prioritized or forgotten.

{{ pt }}
YourMemory

Facts that evolve. Knowledge that merges, decays, and connects across every session — so only what matters survives.

{{ pt }}
Feature
YourMemory
Mem0
Zep
{{ row.f }}
{{ row.ym }}
{{ row.mem0 }}
{{ row.zep }}
// ENTERPRISE[ 06 / 07 ]

Memory across your whole team.

Your team accumulates deep knowledge every day — customer quirks, solved problems, architectural decisions. Right now it evaporates when the session closes. Enterprise gives it a permanent home.

Org Memory shared brain A Alice shared D Dan B Bob shared C Carol E private F private
shared memory · visible to all
private memory · owner only
Without YourMemory

Engineer A spends 2 hours debugging a customer's edge case. Session closes. Knowledge gone. Engineer B picks up the same customer next week — and starts from scratch.

// Session ended context = null customer_hist = null team_knowledge = null
With YourMemory Enterprise

Engineer A solves it once. The memory persists, shared with the org, and surfaces automatically when Engineer B opens a session — solved in minutes.

// Session opened recall("customer edge case") → 3 relevant memories surfaced → last solved: 6 days ago by alice

{{ f.title }}

{{ f.desc }}

Ready to give your team a shared brain?

Enterprise is distributed as a private build. Reach out to get started.

Talk to us about a pilot
// GET STARTED[ 07 / 07 ]

Start in under 2 minutes.

Free, open source, runs entirely on your machine. No API key. No data leaves your system.

Free activation

Get your activation token.

Enter your email — we'll send a 6-digit code to verify.

{{ sendError }}
No password stored · takes 30 seconds
Verify

Enter the code.

Sent to {{ maskedEmail }} — expires in 10 minutes.

{{ verifyError }}
Verified — token ready

Run this command to activate:

yourmemory-register {{ token }}

Token also sent to your email as a backup.

{{ cmd.n }} {{ cmd.title }}
$ {{ cmd.cmd }}

{{ cmd.desc }}

Auto-detects Claude Code, Claude Desktop, Cursor, Windsurf, Cline, Continue & Zed
// FAQFREQUENTLY ASKED

The fine print, in plain English.

{{ faq.a }}

// IN THE PRESSWRITTEN ABOUT
{{ a.source }} {{ a.title }} {{ a.desc }} Read →

Your agent needs
its YourMemory.

Infrastructure for the Agentic Era. Free, open source, self-hostable.

View on GitHub Join Discord